Cover Story: Save Your Brain - More Reasons to Optimize AF Therapy | By Rick McGuire

When it comes to the peripheral effects of atrial fibrillation (AF), the body of evidence keeps growing—so much so you wonder what connection will be seen next. One proving particularly alarming: there appears to be a consistent association between AF and risk of dementia, including Alzheimer’s disease, as well as a suggestion of renal damage if AF is not appropriately controlled. While there may be several mechanisms to explain this association—both AF and dementia share multiple common risk factors, for example—the therapies used to treat AF may influence this risk, too, for better or worse.

In short, type and duration of anticoagulation, as well as the choice between rhythm- and rate-control strategies, might all influence cognitive risk. Also, we’re beginning to better understand the risks of silent cerebral infarcts that are identified in more than 40% of magnetic resonance imaging (MRI) scans of patients with AF.1 Although silent infarcts may not cause acute neurologic deficits, a significant association between silent infarcts and cognitive decline has been emerging.2

Independent Effects

Cardiologists appreciate that a vascular bed is a vascular bed whether it’s in the heart or the brain or situated somewhere else along the peripheral way. It was 18 years ago when investigators reported on a cohort of 952 community-living men, aged 69 to 75 years, in Uppsala, Sweden, showing a strong association between AF and low cognitive function that was independent of stroke, high blood pressure, and diabetes.3 This followed the Rotterdam Study, which first suggested that dementia, Alzheimer’s, and vascular dementia may be related to AF even if no clinical strokes have occurred.

Subsequently, in the United States, Bunch et al. studied 37,025 consecutive patients from the large ongoing prospective Intermountain Heart Collaborative Study database.4 None of the patients had a history of dementia and all patients were followed for at least 5 years. Over the course of the study, 27% developed AF and 4.1% developed dementia. AF was significantly and independently associated with all dementia types (vascular, senile, Alzheimer’s disease, and nonspecified). Although dementia is strongly associated with aging, the highest risk of Alzheimer’s actually manifested in the youngest group, those < 70 years of age (odds ratio: 2.30; p = 0.001). There was a marked increased risk of mortality overall, ranging from a hazard ratio (HR) of 1.38 to 1.45 depending on the type of dementia and all were statistically significant. The greatest risk of death also was in the youngest cohort (< 70 years) with HRs ranging from 1.55 to 2.07.

Further evidence of jeopardy was suggested in a more recent study evaluating the association of prevalent and incident AF with incident dementia in 6,514 participants from the prospective population-based Rotterdam Study; because they were involved early in the investigation of this problem, the study patients were assessed across a 20-year period.5 Risk of dementia was strongly associated with duration of exposure to AF in the younger participants (< 67 years) (in the highest stratum of exposure, HR: 3.30; p  =  0.003 for trend) but not in older participants (> 67 years; HR: 0.25).

But the final jeopardy answer may arise from a new study that examined time to first diagnosis of AF and then time to first diagnosis of any of nine vascular events in a cohort of 4.3 million adults.6 The United Kingdom investigators noted that previous analyses of the relationship between AF and vascular risk have largely focused on stroke, so they used a nationally representative database of health records to look more broadly at the effect of AF on vascular events over a median follow-up of 6.9 years. Baseline AF was associated with a 35% increased risk of subsequent vascular dementia and this combination was associated with a three-fold risk of a fatal vascular event (HR: 3.21; 95% confidence interval [CI]: 1.74 to 5.94).

Repetitive Cerebral Injury

CSWN talked to T. Jared Bunch, MD, FACC, at the recent Heart Rhythm Society meeting in San Francisco, CA. He is director of heart rhythm research at the Intermountain Medical Center Heart Institute in Salt Lake City, UT. Six years ago when his team first reported their evidence suggesting a link between AF and dementia, he said, “At that point, we started to ask the question: Why do we see this association and what are the mechanisms behind it? One of the thoughts we jumped to quickly is that we know AF is associated with macro-events, such as strokes or large bleeds. Perhaps dementia is the micro-event of this. It’s part of a cascade.” Thus, the next question is whether exposure to chronic microembolism or microbleeds results in repetitive cerebral injury that is manifest by cognitive decline. “And with that premise in mind,” Dr. Bunch said, “we thought if that is true, then anticoagulation matters.”

They tested the hypothesis that AF patients with a low percentage of time in therapeutic range (TTR) would be at higher risk for dementia due to under- or over-anticoagulation. Of 2,605 patients, they found that a lower percent of TTR (vs. > 75%) was associated with increased dementia risk: for < 25% the HR was 5.34 (p < 0.0001); for a TTR of 26% to 50%, the HR was 4.10 (p < 0.0001); and for 51% to 75% TTR, the HR was 2.57 (p = 0.001).7

Dr. Bunch said, “The more erratic their warfarin levels—the lower percent of time in therapeutic range—the much higher the risk of dementia. In fact, in multivariate analysis, the strongest predictor of dementia was warfarin anticoagulant irregularity.”

Given that 30%–40% of AF patients are treated with antiplatelet agents due to coexistent vascular disease, he and his team wondered how adding aspirin to the mix might impact risk for dementia, particularly with chronic exposure to over-anticoagulation. The risk was significantly increased in patients using triple antithrombotic therapy.8 Indeed, Dr. Bunch said, “those who were over-anticoagulated and on aspirin had very high dementia rates—on the order of 3%–5% per year.” That led to a study they just presented at HRS.16 seeking to answer a different question: “We asked, is it AF or is it the anticoagulation that is causing the problem?”

They looked at people anticoagulated with warfarin for AF compared to patients being anticoagulated due to thromboembolism, pulmonary embolism, or heart valve disorders. Yes, AF patients were older and tended to have hypertension or diabetes—in other words, more risk factors for cognitive decline. But two things were quite interesting. In their analysis of 10,537 patients managed by the Intermountain Healthcare Clinical Pharmacist Anticoagulation Service who were anticoagulated with warfarin (target international normalized ratio [INR] 2–3) with no history of dementia, AF patients experienced higher rates of total dementia (5.8% vs. 1.6%; p < 0.0001), Alzheimer’s (2.8% vs. 0.9%; p < 0.0001), and vascular dementia (1.0% vs. 0.2%; p < 0.0001) and TTR greatly influenced risk (Figure 1).

However, regardless of the reason why people were being anticoagulated, there was an incremental risk of dementia in people with poor TTR (Figure 2). This was based on a propensity analysis of 6,030 patients to account for differences in baseline demographics. Long-term risk of dementia remained significant in AF patients compared to matched non-AF patients (total dementia: HR = 2.42; p < 0.0001; Alzheimer’s: HR = 2.04; p < 0.0001; senile: HR = 2.46, p < 0.0001) The Kaplan-Meier estimate for survival free of total dementia comparing AF patients chronically anticoagulated versus patients chronically anticoagulated for a non-AF indication can be seen in Figure 3.

Dr. Bunch emphasized that the results are based on a best-case treatment scenario. His team’s anticoagulation strategy is devoted to optimizing warfarin therapy. They educate patients, meet with them regularly, have robo-calls that occur if they don’t show up at the clinic. He said, “Even in this best-case scenario, we are still seeing these dementia patterns and we’re still seeing people who have a lot of volatility with warfarin.”

So when “best case” isn’t good enough, could there be an alternative explanation? Well, elderly people with dementia remain at higher risk of having poor anticoagulation management with more frequent out-of-range INRs. On the other hand, the Intermountain data did suggest the greatest risk was in younger patients. Data from the ACTIVE W (Atrial Fibrillation Clopidogrel Trial With Irbesartan for Prevention of Vascular Events) trial showed that even mild cognitive impairment at baseline is a predictor of less effective anticoagulation (measured by a TTR of < 65% over 1.5 years).9

Indeed, ACTIVE W investigators showed that mild cognitive impairment (as assessed using the Mini-Mental State Examination [MMSE]) was associated with more vascular events (6.7% vs. 3.6% per 100 patient-years; p = 0.002) and more bleeding (9.6% vs. 7% per 100 patient-years; p = 0.04). Therefore, critics have argued that the results of the Intermountain Heart Collaborative Study might be biased by lack of testing to exclude patients with mild cognitive impairment at baseline. However, while bleeding and vascular events were higher in patients with lower MMSE, the risk lost significanceafter adjusting for time in therapeutic range.

What about the non-vitamin K oral anticoagulants (NOACs)? Dr. Bunch and colleagues just reported the results of a retrospective population-based study of 5,254 patients managed with either warfarin or a NOAC (n = 2,627 per group).10

Rivaroxaban was the most commonly used NOAC (55.3%), followed by apixaban (22.5%) and dabigatran (22.2%). Dementia occurred more frequently in patients taking warfarin; however, while statistically significant, the clinical impact of the difference is unclear (0.7% vs. 0.3% of the patients on a NOAC; p = 0.03). Also, they found no difference in rates between the different NOACs. After multivariable adjustment, patients taking NOACs had a 51% decreased risk of dementia incidence or subsequent stroke or transient ischemic attack (TIA) compared with those taking warfarin (HR: 0.49; p < 0.0001).

Dr. Bunch admits there a number of problems with that analysis because of its retrospective nature and the differences in patient characteristics among those who could pay for or had coverage that would cover the NOACs versus those who could only afford warfarin. Consequently, the Utah team has initiated a prospective trial that will randomize people to clinic-managed warfarin therapy versus dabigatran. The study patients will be followed for 2 years with batteries of cognitive function as well as volumetric brain scans so that they can start to see where these events are occurring, the number of such events, and their cumulative impact.

Obviously, the brain is not the only organ along the peripheral pathway, so the Intermountain team looked at long-term renal function and possible correlation to efficacy of warfarin therapy in their AF patients chronically managed in their anticoagulation service. Again, in an almost linear manner, the worse the patient’s warfarin levels, the lower their glomerular filtration rate, the higher their creatinine levels, and the more rapid the progression to renal failure.

“No matter which metric we looked at,” Dr. Bunch said, “whether they started with normal renal function, mild or moderate dysfunction, if their anticoagulation control was poor, they progressed—in a similar analogy to the brain study but looking at a different organ: the kidneys.”

Patient X

The initial patient who prompted the Intermountain group’s study into the association of dementia and AF would speak coherently while in sinus rhythm. However, when his rhythm would change to AF, he would abruptly lose his thought process and look to his wife to complete his sentence. While the patient would refer to these as “senior moments,” Dr. Bunch could see the heart monitor behind the patient and the effect was clear every time he slipped into AF or back into sinus rhythm. “I started asking other doctors and they would say, ‘I’ve had patients like that,’ and that began the whole idea of research into cognitive function and AF.”

That’s not explainable by clots or bleeds, so what might the mechanism be? An autopsy study of 84 Alzheimer’s patients showed the most common form of cerebral vascular atherosclerosis in these patients was in the Circle of Willis, which is critical for cerebral perfusion. It creates a means of redundancies and collaterals that will provide generalized blood flow throughout the brain, even if more proximal atherosclerosis develops. Thus, it is likely that people with Circle of Willis atherosclerosis, when exposed to beat-by-beat alterations in blood flow, would not retain the same ability to distribute cerebral blood flow globally as those without atherosclerosis. This might explain a patient with abrupt cognitive dysfunction upon onset of AF.

This whole issue brings us around to a controversy that seemed to have ended years ago in terms of which is better: a rate- or a rhythm-control strategy. The issue seemed to be settled by the AFFIRM (Atrial Fibrillation Follow-up Investigation of Rhythm Management) study, which suggested rhythm control offers no survival advantage over rate-control strategy, plus there may be a lower risk of adverse drug effects with a rate-control strategy.11 However, mean age of study participants was almost 70 (mean: 69.7 years) and mean follow-up was 3.5 years. More recent data suggest that patients treated with rhythm control had reduced mortality when follow-up was extended beyond 4 years.12

But even 4 years is a drop in the bucket compared to a decade or more with AF. Eric N. Prystowsky, MD, FACC, is a pioneer in electrophysiology and heads up the cardiovascular disease program at St. Vincent Hospital in Indianapolis. At an AHA meeting, he noted that there are no safety data related to allowing patients to stay in AF for 20, 30, or 40 years. Given evidence of potential deleterious effects of longstanding AF, such as cognitive effects, dementia, and even Alzheimer’s, Dr. Prystowsky said, “I am not telling you what to do. What I am asking people to do is to rethink this issue with your patients. There are new data suggesting that leaving patients in A-fib is not as safe as you think.”

According to Dr. Prystowsky, the “prime directive of AF management” is to preserve the brain. He has written about this in JACC,13 emphasizing that stroke is clearly not the only neurological consequence of AF. He mentioned the data pointing to cognitive impairment and silent cerebral infarcts in patients with AF. In what appeared to be the first study to investigate the relationship between AF and cognitive decline based on cerebral MRI, investigators reported that patients with either paroxysmal or persistent AF have a higher prevalence of silent cerebral ischemia and worse cognitive performance compared with controls.2 “This is a passion of mine,” said Dr. Prystowsky, “and I think it is, unfortunately, not something we always think about as much as we should.”

CSWN asked Gregory YH Lip, MD, FACC, director of the haemostasis thrombosis and vascular biology unit at the University of Birminham in the U.K., for his opinion. Given the growing evidence of decreased cognitive function and even dementia, even deleterious effects on renal function, especially when AF is poorly controlled, is it time to go beyond stroke and consider other peripheral effects of AF when choosing therapy? He agreed, adding “We forget that AF independently increases all-cause mortality—even in the historical placebo-controlled trials, warfarin reduced stroke by 64% and all-cause mortality by 26%.” Optimal AF management leads to reduced risk across the board, but studies show that too many patients aren’t receiving anything close to optimal care, with some clinicians still mistakingly using aspirin as some sort of “safer alternative” when, in fact, Dr. Lip noted aspirin has no impact on mortality. None. Even worse: it’s not just that aspirin is no substitute for antithrombotic therapy. He pointed to a paper form his group that is online before print demonstrating that there is no net clinical benefit for antiplatelet therapy compared to no antithrombotic therapy at all.14

Risk Aversion

If you’re trying to reduce cognitive risks, one answer is certainly NOT to avoid anticoagulation in patients who qualify for such therapy. Remember that UK study of 4.3 million people cited above? For each of the nine vascular events analyzed, the investigators evaluated risk with and without antithrombotic therapy. Baseline AF was associated with a 57% higher risk of vascular dementia (HR: 1.57; 95% CI: 1.14–2.17) with no antithrombotic therapy; for those who were on antithrombotic therapy, the risk was only 29% higher (HR: 1.29; 95% CI: 1.08–1.54). Considering only fatal events, the comparable risk was 4.91 (95% CI: 1.46–16.48) and 2.88 (95% CI: 1.44–5.78), respectively.

Looking more broadly, AF with baseline antithrombotic therapy was associated with a 15% higher risk of any vascular event (HR: 1.15; 95% CI: 1.12–1.18) and a 69% higher risk of a fatal vascular event (HR: 1.69; 95% CI: 1.61–1.78). However, AF without antithrombotic therapy was associated with two times the risk of any vascular event (HR: 2.15; 95% CI: 2.05 to 2.24) and two-and-a-half times the risk of a fatal vascular event (HR: 2.64; 95% CI: 2.43–2.86).

This topic has been of intense interest to A. John Camm, MD, FACC, head of the Department of Cardiac and Vascular Sciences at St. George’s Hospital at the University of London, UK. Dr. Camm, who is also the first author of the European Society of Cardiology AF guidelines,15 said physicians are more concerned about the strokes their therapy might cause than the strokes likely to be prevented with proper anticoagulation. However, the increased risk of hemorrhagic stroke in a patient undergoing appropriate anticoagulation is roughly one or two per 1,000 cases. Conversely, risk of thromboembolic stroke in a patient who is not on anticoagulant therapy is nearer 20 per 1,000. So, Dr. Camm stresses that there is approximately a 10-fold difference between the strokes prevented by guideline-recommended anticoagulation versus serious strokes caused by such therapy.

When asked by CSWN, he too agreed that “Stroke is only one effect of AF on the brain—another is dementia. AF also increases hospitalizations, increases heart failure, increases mortality, etc. Thus, preventing stroke is only one aspect of AF care, but it is a very important element of care.”

Indeed, if a University of Minnesota study is correct, the cognitive decline in people with AF is mediated by subclinical cerebral infarcts (SCIs) as detected by magnetic resonance imaging. The study was led by Lin Yee Chen, MD, FACC, a University of Minnesota cardiologist who analyzed 935 participants of the Atherosclerosis Risk in Communities Study.16 Among participants without SCIs on MRI scans, incident AF was not associated with cognitive decline. In contrast, incident AF was associated with greater annual average rate of decline in word fluency (p = 0.002) among participants with prevalent SCIs “Our findings have important clinical and public health implications,” he said. “They raise the possibility of anticoagulation as a strategy to prevent cognitive decline in patients with AF.”

Protecting the Brain

Despite available data, the need looms large for sizable multicenter studies to examine the effects of NOACs, rhythm and rate control, and left atrial appendage occlusion on long-term cognitive function. Heart rate may be important because Bunch et al. have found a linear, and inverse risk association with heart rate and dementia with lower heart rates increasing dementia risk. The Utah team is planning a randomized trial of pacing rates and their effect on cognition. Also, they will do further work in evaluating a genetic marker that tracks for early-onset AF and stroke and those patients who have very high dementia rates.

Until large multicenter and randomized clinical trial data are completed, what can be done now to save the brain?

According to Dr. Prystowsky, one way to lower stroke risk and better protect the brain: assess risk using the broader CHA2DS2-VASc risk assessment tool (a Class I recommendation in the current ACC/AHA guidelines).17 With prior stroke, TIA, or a CHA2DS2-VASc score > 2, oral anticoagulants are recommended (Class I recommendation) based on shared decision making, with a discussion of the risk of stroke and bleeding that considers patient preferences (also a Class I recommendation). “In my own experience,” said Dr. Prystowsky, “it takes about 10 minutes to go over this with my patients. It’s a very important decision that is well worth the time and then you make a shared decision.”

Dr. Bunch said he is trying to reinforce that warfarin is safe if it’s managed well. As a participant in the pivotal trial for rivaroxaban and a member of that trial’s stroke advisory committee, Dr. Bunch said some clinics put people on warfarin for the first time and then didn’t check their INR until 4–5 weeks later. That approach, he said, is “exposing people to crazy ranges of therapeutic. As a community that uses warfarin, we have to get away from those practices.”

Catheter ablation is emerging as an effective alternative to maintain patients in sinus rhythm. Short term, AF ablation seems to increase postoperative cognitive dysfunction; it’s detectable in 24% of patients at day 2 and 13% at 90 days.18 However, in the longer term, ablation shows promise for decreasing the risk of all types of dementia.19 Besides maintaining sinus rhythm and oral anticoagulation, there is also evidence that aggressive risk factor modification may reduce the likelihood or delay the onset of dementia.20

Dr. Bunch said the risk for dementia among all AF patients is lowest among his patients treated with catheter ablation who maintain sinus rhythm long term. Yes, the low rates of dementia likely reflect healthier patients selected for ablation, the complex system of catheter ablation, and resulting follow-up care. Nevertheless, Dr. Bunch suspect that it also reflects cessation of long-term anticoagulation exposure, improving cardiac output, and minimizing beat-to-beat variability in some low-risk patients.

Dr. Bunch said he and his team get e-mails from all over the world wondering about effective therapeutic strategies and their first message is always, “If you have good control (with warfarin), you’re in a good spot—don’t stop.” But if a patient’s INR is erratic then, he said, “let’s think of some of the alternatives. It may not be the Watchman device—that approach is for one type of stroke but it leaves you vulnerable to others. However, at least it gives you incentive to go down those roads and ask those questions.”

The mention of the Watchman™ device for left atrial appendage closure was not by happenstance: the prospective study from his team will also be incorporating a Watchman™ arm, and the study additionally includes a comparison of dabigatran versus warfarin. Hopefully, he said, they will have data from all those arms in about 3 years.

In the meantime, what we know—and hopefully won’t forget—is the clear relationship between AF, dementia, and other markers of cognitive deficit. Fidelity to effective anticoagulation and antiarrhythmia strategies can go a long way to saving the heart—and brain.


  1. Kalantarian S, et al. Ann Intern Med. 2014;161:650-8.
  2. Gaita F, et al. J Am Coll Cardiol. 2013;62:1990-7.
  3. Kilander L, et al. Stroke. 1998;29:1816-20.
  4. Bunch TJ, et al. Heart Rhythm. 2010;7:433-7.
  5. de Bruijn RF, et al. JAMA Neurol. 2015;72:1288-94.
  6. Emdin CA, et al. Int J Epidemiol. 2016 May 3. [Epub ahead of print]
  7. Jacobs V, et al. Heart Rhythm. 2014;11:2206-13.
  8. Jacobs V, et al. J Cardiovasc Electrophysiol. 2015 Aug 13. [Epub ahead of print]
  9. Flaker GC, et al. Circ Cardiovasc Qual Outcomes. 2010;3:277-83.
  10. Jacobs V, et al. Am J Cardiol. 2016 May 5. [Epub ahead of print]
  11. Wyse DG, et al. N Engl J Med. 2002;347:1825-33.
  12. Ionescu-Ittu R, et al. Arch Intern Med. 2012;172:997-1004.
  13. Prystowsky EN, et al. J Am Coll Cardiol. 2013;62:540-2.
  14. Fauchier L, et al. Stroke. 2016 May 26. [Epub ahead of print]
  15. Camm AJ, et al. Eur Heart J. 2012;33:2719-47.
  16. Chen LY, et al. Stroke. 2014;45:2568-74.
  17. January CT, et al. J Am Coll Cardiol. 2013;62:531-9.
  18. Medi C, et al. J Am Coll Cardiol. 2013;62:531-9.
  19. Bunch TJ, et al. J Cardiovasc Electrophysiol. 2011;22:839-45.
  20. Kanmanthareddy A, et al. Curr Cardiol Rep. 2014;16:519.
Read the full July issue of CardioSource WorldNews at

Clinical Topics: Arrhythmias and Clinical EP, Noninvasive Imaging, Atrial Fibrillation/Supraventricular Arrhythmias, Magnetic Resonance Imaging, Sleep Apnea

Keywords: CardioSource WorldNews, Alzheimer Disease, Atrial Fibrillation, Cognition, Cognition Disorders, Dementia, Magnetic Resonance Imaging

< Back to Listings